High-affinity binding of interferon-gamma to a
basement membrane complex (matrigel).
H Lortat-Jacob, … , H K Kleinman, J A Grimaud
J Clin Invest.
1991;
87(3)
:878-883.
https://doi.org/10.1172/JCI115093
.
Recently it was demonstrated that growth factors are bound to the extracellular matrix, and
can regulate cell behavior. Using three different types of binding assays, we have examined
the interaction of interferon-gamma with a basement membrane produced by the
Engelbreth-Holm-Swarm tumor. Basement membrane was found to bind interferon-gamma
in both a time- and concentration-dependent manner. Equilibrium binding analysis revealed
a high-affinity site with a dissociation constant of 1.5 10(-9) M and a maximum binding
capacity of 1.6 10(9) sites/mm2 of basement membrane. Competition studies show that the
binding is inhibited by heparan sulfate, suggesting that basement membrane-heparan
sulfate proteoglycan could be the binding site. This interaction was clearly confirmed by
native polyacrylamide gel electrophoresis and dot-blot analysis with purified basement
membrane molecules. Furthermore, the carboxy-terminal part of the interferon-gamma
molecule contains an amino acid cluster, very closely related to a consensus sequence,
present in more than 20 proteins known to bind sulfated glycosaminoglycans such as
heparin. These data demonstrate a possible role of extracellular matrix components in
storing cytokines and in modulating the cellular response to such factors.
Research Article
High-Affinity Binding of Interferon-ey
to
aBasement Membrane
Complex (Matrigel)
Hugues Lortat-Jacob, Hynda K. Kleinman,* and Jean-Alexis Grimaud
Centre National delaRechercheScientifique URA602,InstitutPasteur, Lyon,France;and*Laboratory ofDevelopmentalBiology
andAnomalies, National InstituteofDentalResearch, National InstitutesofHealth, Bethesda, Maryland20892
Abstract
Recently it was demonstrated that growth factors are bound to the extracellularmatrix,and can regulate cell behavior. Using threedifferenttypesof binding assays, we have examined the interaction of
interferon-t
with a basement membrane pro-ducedby the Engelbreth-Holm-Swarm tumor. Basement mem-brane wasfoundtobind interferon-y in both a time- andcon-centration-dependentmanner.Equilibrium binding analysis re-vealed ahigh-affinity site with a dissociation constant of 1.5
10-'
Mand amaximum binding capacity of 1.610'
sites/mm2 of basement membrane. Competition studies show that the bind-ing is inhibited by heparan sulfate, suggestbind-ing that basement membrane-heparan sulfate proteoglycan could be the bindingsite.Thisinteractionwasclearlyconfirmedbynative polyacryl-amide gel electrophoresis and dot-blotanalysis with purified basement membrane molecules. Furthermore, the carboxy-ter-minal part of the
interferon-'y
moleculecontains an amino acid cluster, very closely relatedto aconsensus sequence, present in morethan20 proteins known to bind sulfatedglycosaminogly-canssuch asheparin.These datademonstrate a possible role of extracellular matrix components in storing cytokines and in modulating the cellular response to such factors. (J. Clin.
In-vest. 1991.87:878-883.). Key words:bindingsite * extracellu-larmatrix * local concentration * heparan sulfate proteoglycan
Introduction
Increasing evidence nowdemonstrates that cellularactivities
depend on multifactorial events involvingsoluble-phase
fac-torssuchascytokines (1, 2),and/orsolid-phase factorssuch as extracellularmatrixcomponents(3,4). These soluble-or solid-phase factors bind to specific cell surface receptors, some of which have been characterized (5, 6). Basement membranes
arespecializedextracellularstructureswhich underlie epithelia
andarecomposed of collagen IV, laminin, nidogen/entactin,
and proteoglycans. Thesematricesaffectavariety of biological activitiesincludingcelladhesion(7,8),migration(8, 9), growth and differentiation, neurite outgrowth
(10),
regulation ofcellmorphology
(1
1),tumormetastases(6, 12-14), and regulation ofmorphogeneticprocessesduringdevelopment(15).The cel-lular response depends on the cell type and amount andformofthebasement membrane. Recently, it has been shown that allfourcomponents both alone and incombinationareableto exertbiological activity(8, 16-21). Laminin appears to be the
AddressreprintrequeststoDr.Grimaud,UnitedePathologie Cellu-laire, CNRS URA 602, Institut Pasteur, Avenue T.Garnier, 69365 Lyon Cedex 7,France.
Received.for
publication9May1990andinrevisedform14August1990.
mostactive in this regard (10, 17, 20). The large heparan sulfate proteoglycan termed pearlecan has recently been cloned and
partially sequenced andcontainssome40%homology to the amino-terminal end of the laminin chains (21). It has been shown to promote hepatocyte adhesion and a specific receptor hasbeenisolated (22). This, however, does not appear to be the only function of the heparan sulfate which also servesto regu-late the passage of macromolecules through the basement membraneowing to its strongnegative charge (23).
Cytokines are soluble factors, with multiple overlapping cell regulatory actions (1). Theresponse ofa cell to a given
cytokineisdependent onthe local concentration of the cyto-kine, the cell type, and other cell regulators to whichit is
con-comitantly exposed (1).Forexample,
interferon-y
(IFN-"y)
isapleiotropic factorthat has manyeffectsondifferent cells (24).It
inducestheantiviralstate inuninfectedcells, can regulate cell growth and differentiation, regulates the immune response, and inhibits tumor cell growth (24-27). Although cytokines
suchas
IFN-'y
aregenerally thought ofassolublefactors, itispossiblethattheseproteinsmay be inabound form, and thus, either be stabilizedtoprevent breakdownand/or immobilized to allow continued and localized stimulation ofcells.Recently, growth factorssuch as fibroblastgrowth factor (FGF),' have been found to be incorporated into the extracellular matrix where they are stored. In the caseof FGF, the thin extracellular
matrix ofthebasement membrane wasfoundtobeastorage depot and the major ligand washeparan sulfate
proteogly-can(28).
Here wehave investigatedwhether
IFN-"y
canbindto the basementmembrane. Since intactbasement membranefrom tissues is difficulttoisolatereproduciblyand inpurity,wehave usedareconstituedbasementmembrane from theEngelbreth-Holm-Swarm(EHS)tumor tostudytheinteractionof IFN-y. This basementmembrane matrix, termed matrigel, contains
allofthe componentsof authenticbasementmembrane and in the electronmicroscope hassimilar lamina densa-like
seg-ments to those observed in tissue-derived basement
mem-branes(29-3 1).
Methods
1251I
iodinationofIFN-"y.
Recombinant humanIFN-'y
withaspecificactivity of 2.10' U/mgwasa generousgift of Roussel Uclaf, Paris,
France. 10jgof IFN--ywasreacted in 50
Al
of 0.2Mphosphatebuffer,pH 7.4,containing 0.8MNaCl,0.02% Tween20,500 MCi ofNa125I
(NewEngland Nuclear-DuPont, Boston, MA),and2.5jigof
chlora-mineT.AfterIminatambienttemperature,the reactionwasstopped
by addition of 10
AI
of sodium metabisulfiteat0.3 mg/ml. Proteins were thenseparated from free iodine by molecularsievingon apre-packedSephadexG25PD 10column(Pharmacia-LKB, Uppsala,
Sweden),equilibratedwith0.1Mphosphate buffer,pH 7.4, containing
The JournalofClinicalInvestigation,Inc.
Volume87, March 1991, 878-883
1.Abbreviations usedinthis paper: FGF, fibroblast growthfactor; GM-CSF, granulocyte/macrophagecolony-stimulatingfactor.
0.5 M NaCI, 0.02% Tween 20, and 10% of fetal calf serum. 251I-labeled
IFN-y,of which thespecificradioactivity was 20uCi/ug, was aliquoted and stored at-20°C. The high ionic strength buffer used here was foundtobe necessary topreventaggregation ofIFN--y.If labeled in more physiological conditions(e.g., PBS), IFN-'y eluted in the void volume of a G75 Sephadex column, as reported by others (32).
Preparationof matrigel-coated dish. Matrigel wasprepared as de-scribed (29). This mixture of matrix proteins is in solutionat4°C but forms a gel when warmedat24-37°C. Matrigel was thawedat4°C and diluted to 5 mg/ml with ice-cold 0.15MNaCl, 50mMTris-HCI, pH 7.4, and 30 ul per wellwasallowed to gel overnightat35°C ina96-flat botton well assay plate (Pro Bind 3915, Becton, Dickinson & Co., Lin-coln Park, NJ). Wellswerethensaturated for 1 h at 37°C with 200
,l
of 50 mMTris-HCI, pH 7.4,containing0. 15MNaCl and 20mg/ml BSA.Binding studies.Ingeneral, 100 LIof unlabeledIFN-y(12.5ng/ml
to5
gg/ml),
containing25nCiof radiolabeled IFN-yemployedas atracer,wasincubated inmatrigel-coated wells,at roomtemperature. After the time necessary to reach equilibrium (or at different times for kineticstudies),thesupernatant fractionwasremoved and thematrigel
was rapidly washed with 200 Ml of PBS/BSA 0. 1%/Tween 20 0.02%. Bound radioactivitywasthen removedbydissolvingthematrigel with
4 M NaOH. The supernatant fraction plus the wash and dissolved matrigelwerecounted inagamma counter(1260 Multigamma, Phar-macia-LKB) to determine free and bound radioactivity. Experiments were performed twice, with triplicates for each data point. The quantity ofIFN-'ybound in each wellwascalculated from thespecific radioactiv-ity ofIFN-,yin each concentration. Nonspecificbinding was deter-mined according to Chamness and McGuire (33) using a lim(B/F)
= 0.2. To estimate Kd and Br,,,, (maximum binding capacity), data were represented as specific bound ligand plotted against thelogarithm
of free ligand concentration, according to the equation
B.
=M,
*Kd
.F/(
I +Kd*
F),
inwhichB,andM,are"surfaceconcentra-tions" ofbound IFN-y and coated matrigel, respectively, as de-scribed (34).
Enzymaticdigestion of heparansulfate. Matrigelwaspreparedas
described above and incubatedovernightwith50 Ml ofCaCl2(IMuM),
Tris-HCl(50 mM)containingeither 10 U ofheparitinase (Miles Scien-tificDiv.,Naperville, IL)or2.5 units ofheparinase (SigmaChemical Co., St.Louis, MO).Afterwashing, 100,ul of'2511IFN--y(25nCi)was
incubated for4 hwith the matrigel. Boundradioactivitywas deter-mined. Nonspecific bindingwascalculated with a200M excessof cold IFN.
Competitionstudy. Matrigel-coatedwellswereincubated (4 hat roomtemperature) with 100Mlof radiolabeled IFN-y(25 nCi/ml),in the presence ofincreasingconcentrations ofcompetitors (heparinfrom Serva,Heidelberg, FRG; heparansulfate fromSigmaChemicalCo.;
andrabbit anti-humanIFN-yfromGenzyme Co.,Boston,MA).
Non-specific bindingwasdetermined with a200 Mexcess of cold IFN. Boundradioactivitywasthendeterminedasdescribed above and the percentage ofbindinginhibitionwascalculated.
Nativepolyacrylamidegel electrophoresis(PAGE).This electropho-retic systemwasdevelopedtomonitor interactions between proteins that have different isoelectricpoints.The 5%acrylamide/0.5%bis acryl-amidegel,in 50mMTris/40mMboric acid, pH 8.5,wasallowedto
polymerizeasusual undervacuuminahorizontal gel mold. The comb was inserted in the middle of the gel.
Samples consisted ofIFN-y
(5Mgg)
aloneorinmixture with different matrix componentssuchaslaminin (5gg,GibcoLaboratories, Grand Island, NY), collagen typeIV(5Mg,
Gibco Laboratories), heparan sul-fate (2.5 ag),andmatrigel (12.5 Mg). Beforeelectrophoresis, sampleswereincubated4 hat4°C. Tank buffer was Tris/borate buffer, pH 8.5.
Running conditionswereconstant voltage (200 V) for I h at 4°C. Proteinswererevealedby Coomassie Blue R-250 staining as usual.
Dot blots.Adescribed procedurewasfollowed to prevent glycosami-noglycans loss (35). Briefly, I ML ofglycosaminoglycan (from I to 10 mg/ml in PBS) was appliedonto anitrocellulose sheet and air-dried. Thenitrocellulosewasthen fixed in5%paraformaldehyde + 0.5% ce-tylpyridinium chloride for 30min at roomtemperature and washed in
buffer. Purified basement membraneproteins (10
Mg)
wereappliedontonitrocellulosebysuctionasusual. Thenonoccupied bindingsites
wereblocked withPBS/BSA3% for 30 minat37°C.Thenitrocellulose
stripswereincubatedovernightat4°Cwithinterferon-y
(1I
MinPBS/BSA 1%),washed inPBS,andincubated with rabbitanti-IFN-y fol-lowedbyperoxidase-labeledanti-rabbitIgG. Alternatively radiola-beledIFNwasused,andafterwashing,thenitrocellulosefilterswere
directlycounted.
Domainmapping with monoclonal antibodies. MAbsB22, B27,32,
and35 which define differentepitopesonthe IFN-y moleculeare
de-scribed elsewhere(36).Radiolabeled IFN--ywasallowedtoreact, inthe
presence of differentMAbs,withheparansulfate
(1I
g),usingthe dot-blot assay. Afterwashing,each nitrocellulosestripwascountedtode-termine the boundradioactivity.
Interferonassay. Antiviralactivity wasdetermined with a micro-titer inhibition ofcytopathiceffect assay
against
vesicularstomatitis virus(VSV) on amonolayerof Wish cells. Cells(5 x104/well)weregrown in 100MlofDMEsupplementedwith 10%fetal calfserum on
plasticor on basement membrane(30
Mig/well).
Before cellculture,basement membraneswereincubated 6hat4°CwithIFN-yand
exten-sivelywashed.Then,24 hafter the addition of thevirus,cell
viability
wasdetermined with the3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide assayasdescribed
(37).
Inparallel experiments,
IFN-vy
wasusedtodetermine theexactquantity
ofIFNremaining
on thebasementmembranejust
beforecellculture.Results
Binding of
IFN-,y
towhole basement membrane. Whenaddedto a
layer
ofgelled
basementmembrane,
radiolabeledIFN-y bindsinatime-dependent
manner(Fig.
1).
Atroomtempera-ture,the
steady
stateisreached in 3.5 h. Under thesamecondi-tions, binding
ofIFNtoplastic
(noncoated wells)
wasfoundtobe nonexistent.
Scatchard
analysis
ofIFN--y
binding
to basementmem-branewascarriedout
(Fig.
2a).
Thebiphasic
curveindicatedanonspecific
binding
(horizontal
part oftheplot)
and asingle
classof
binding
sites.Thedatawerecorrectedforspecific
bind-ing
asdescribed(33),
andplotted
withthesemilogarithmic
rep-resentation
(38)
(Fig.
2b).
From thes-shape
curve,representing
the
binding site,
the dissociation constantKd
= 1.5 nM wasdetermined. A
high affinity
between IFN-y and basementmembranewasdemonstrated.The numberof
binding
siteswascalculatedat 1.6
109/mm2
ofcoatedbasementmembrane.Binding
inhibition assay.Wenextstudied thenatureoftheinteractionbetween IFN-y and basement membrane
(Fig.
3).
IFN-y was incubatedon
matrigel
inthe presence ofdifferentcomponents, and the inhibition of
specific binding
wasmea-sured.NaCl
(3 M),
butnotTween 20(0.5%),
inhibited 100% ofthe
binding,
suggesting
that the type of interaction ismoreionicthan
hydrophobic.
As acontrol,
polyclonal antibody
05
t "o
z 0
0
O i
Figure 1.Effectof time
on thebinding ofIFN--y
0 @ tobasementmembrane.
Basement
membrane-coated(e)ornoncoated
went m9mbrene *---- wells (o) wereincubated
IC c-O at room temperature
with 1.25 ngof radiolabeled IFN-y.
Afterthetimeindicated,
I
0~~~~~
2
z
a
z
0
0 25 50 75
BOUND IFN(ng)
1 Bmax/2=0.62sng
b
_ , d = i.snM0 0.3 3
FREE FN ( nM)
Figure 2. Effect ofIFN-yconcentration
onthebindingofIFN-,ytobasement membrane. 25nCi/mIof radiolabeled
IFN-'ywasincubated for 4 h at room temperature, with 150,ugof coated basement membranecomplexand with increasing concentrations of coldIFN-,y
/ (from 12.5 ng/ml to 5 ug/ml). Bound andfreeIFN-,ywerethen measuredas
described in Methods.(a)Scatchard plot of the raw binding data. (b) Semi-logarithmic plot of the data corrected
30 for nonsaturable binding, using 0.2as
thelimit of bound/free.
against IFN-y (diluted 1/300)wasfound to fully preventthe
specific binding. Glycosaminoglycans such asheparan sulfate orheparin (20 ug/ml) inhibited thebindingtothe same extent.
Binding ofIFN--y to basement membrane maytherefore be due to the componentheparansulfate proteoglycans.
NativePAGEanalysisofthe binding. Electrophoresisunder native conditions was carried out to further investigate this binding, (Fig. 4). In this system,
IFN-y,
ofwhich the pI=8.7, isslightly positive andwillmigratetowardthe cathode (lane 1). Incontrast, BSA (pl=5.2)isnegatively charged and migrated toward the anode(lane 2). When
IFN-y
andBSA aremixed, theirmigration is delayed(becauseof opposites, charges whichmayresultin someinteraction),but eachprotein migratestoits
side(lane 3). Heparan sulfate glycosaminoglycan, which is highly negative, should migrate (lane6a)toward the anode but cannotbe visualized because it is notstained by Coomassie Blue. When
IFN-y
is electrophoresed in thepresenceof hep-aransulfate (lane 6b), it migratestowardthe anodeinstead of thecathode, demonstrating thatthere isa strong interaction between IFN-'y and heparan sulfate. 2.5 Mg of heparan sulfatewasable to completely bind IFN-y (5 ,g) and to permit its
migration
totheglycosaminoglycan side.The same experimentwas performed with laminin (lane
4a) and laminin plusIFN(lane
4b)
orcollagentype IV(lane5a)andcollagen typeIV plus IFN (lane5b). In allcases,the migration of IFN-,y is notsignificatively affected by the pres-enceof either lamininorcollagen type IV, showingthatthere
are no interactions between these components. Whole
base-mentmembrane complex (matrigel) and matrigel plus
IFN-'y
were also tested with two different preparations of matrigel (lanes 7 and 8). Only a small part of theIFN-'ymigrates on its side(cathode side) confirming that matrigelbindsIFN
(com-parecathode side oflane 1 with cathode side oflane 7b and 8b). Ascontrol, ribonuclease A (Sigma Chemical Co.), which is also basic-pI = 8.9 (lane 9a)-and ribonuclease A plus heparan sulfate(lane 9b)weretested, and ribonuclease migrationwas
notaffected by the presence ofheparansulfate.Thus,the inter-feron-heparan sulfate interaction appears specific.
Cathode side
-1 2
3
4
5 6a
b
ab
ab
7
a
b
8
9
a
b
a b_
-"_
____-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...
NaCI(H) Twin 2o(VI) HP(pg/ml)
INHIBITORS
HS(pg/ml) Ab(dilution)
Figure3. Effect of various inhibitorsonthebindingof IFN-yto
basement membrane.
IFN-'y
wascoincubated withNaCl,Tween20, heparin(HP), heparan sulfate (HS),orpolyclonalantibody againstIFN--y diluted in PBS/BSA 1%(Ab), 4h at roomtemperature. Nonspecific binding was determinedastheboundradioactivity in the presence ofa200M excessof cold IFN-y.
U
Anode
side
..._.p.:AO
Figure4.Effect of isolatedproteinsof the basementmembrane,and whole basement membrane complex,ontheIFN-,ymigrationina
native PAGE systematpH=8.5. Lane1,IFN(5gg). Lane2, BSA
(5 Mg).Lane3,IFN(5Mg)+BSA(5 Mug).Lane 4a, laminin(5 ag);4b, laminin (5,gg)+IFN(5pg).Lane5a,collagentype IV(5ag);5b, collagen typeIV(5 Ag)+IFN(5ag).Lane6a,heparansulfate(2.5
Ag); 6b, heparansulfate(2.5 ug) +IFN(5 ag).Lane 7a and 8a, matrigel(12.5ag), 7b and8b, Matrigel (12.5 ug)+IFN(5ug).Lane 9a, ribonucleaseA(5ag), 9b,ribonucleaseA(5Ag)+heparansulfate
(2.5
ug).
a4
.
a
z
0 U 0.3
a2
10
N
z
I-5 rc
0 0
liv-w- CMgo
I
I9
3!
IMIR
RI
U)IS!
RI
% % %Enzymaticdegradation ofbasement membrane-glycosami-noglycan. Matrigel was incubated with either heparinase or heparitinase. Radiolabeled
IFN-y
wasallowed to react as de-scribed above with enzyme-treated matrigel orwith untreated matrigel (to determine the 100% of binding). Digestion with heparinase completely abolished the binding, while digestion with heparitinase reduced it to 50% only. These data suggest that IFN binds to heparan sulfate heparin-like sequences, but we cannot rule out the fact that heparinase digestion may be more complete than that of the heparitinase.Binding ofIFN--y topurified basement membrane
compo-nents. In order to measure the direct binding ofIFN-'yto puri-fied basement membrane components, ligand dot-blot experi-ments wereperformed. As shown in Fig. 5, only heparan
sul-fatebindsIFN-,y, confirming the binding activityfound inthe Scatchardanalysis. Whole basement membranecomplex(10
gg)
dot-blottedontonitrocellulose filters also bindsIFN-Y,butweakly, presumably becauseof the lowcontentof heparan sul-fate(30).Theexperimentwasalso carriedoutwith other
extra-cellularmatrix components including collagen type I, III, and fibronectin. Noneofthese components boundIFN-y(data not shown).
In order to assess the specificity between heparan sulfate andIFN-y, radiolabeledIFNwas also allowed tointeractwith different glycosaminoglycans. As shown in Table I, keratan
sulfate,dermatansulfate, and chondroitin sulfatedo not
signifi-cantly bindIFN-y.Thisbindingseems tobe restricted to
hepa-rinand to heparan sulfate(the latter containing heparin-like sequences). The exactpolysaccharide residues involved in the
bindingremaintobedetermined.
Domainmapping. IFN-'ywasallowedto reactwith heparan
sulfate, in the presence ofMAbsto IFN--y. Among the four
MAbs tested, only B22 prevent thebindingbetween IFNand heparan sulfate (Table II). This MAb recognizes the last 15
amino-acids (carboxy-terminal side) ofthe
IFN-'y
molecule, whereas B27, 32, and 35 define other domains (36). These data demonstrate that thecarboxy-terminal partofIFN-,y is in-volved in its binding to heparan sulfate.Antiviralactivity
of
basement membrane-bound interferon.Basement membrane-coated wells were incubated with various
concentrations of
IFN-'y,
washed,and testedfor antiviral activ-ity.Asshown(Table III), cells grown on basement membrane-bound IFNwereprotected against thecytopathiceffect of thevirusin a dose-dependent manner, whereas cells grown on base-mentmembrane alone were not. Thus, the IFN--y bound to the basementmembrane isbiologically active.
*
_a
Figure 5. Dot-blot analysis of the binding of
IFN--y.IFN--y(1 MM)was incubated with
_.. cb
nitrocellulosestrips
dot-blottedwith 10 Ag of(a) heparan sulfate,(b)collagen type IV, (c) laminin, and(d)whole basement membranecomplex. After washing,IFN--y_ wasstained with rabbit anti-humanIFN--y
-* U followedby peroxidaselabeledanti-rabbit IgG.
Table L. Interactions ofIFN-yandGlycosaminoglycans
Glycosaminoglycandot-blotted Nitrocellulose bound onto nitrocellulose filter radioactivity
cpm
Keratansulfate 117.1
Dermatansulfate 1,254.0 Chondroitin sulfate 2,140.6 Heparansulfate 13,438.5
Heparin 28,224.2
Radiolabeled IFN-ywasincubatedovernightat40C withthe indi-catedglycosaminoglycans (1zg) dot-blottedontonitrocellulose. After washingin PBS, the nitrocellulose strips were counted.
Discussion
Previous datareported thatextracellular matrix components may interact with growth factors such as FGF (28) or with
cytokines suchasIL-3or granulocyte/macrophage colony-stimulating factor (GM-CSF) (39). We present, for the first
time, the binding of human interferon--y to basement
mem-brane, asdetermined in several differentand specific binding
assays. Bindingdatarevealahigh-affinitysite forIFN--ywitha
Kdof1.5 x 10-9 M anda maximum binding capacity of 1.6
x
10-'
IFNmolecules/mm2
of basement membrane.A relatively long time is needed for IFN-y binding. This couldbeexplained by the three-dimensionalstructure of the
basementmembranewhich could slow diffusion of molecules. Itshould be also noted that the heparansulfateproteoglycan
represents < 5% ofthe total protein present in matrigel (30). Thisalsoexplainstheimportance of nonspecific binding(B/F
= 0.2 in the Scatchard plot) which corresponds to IFN--y trapped, but not bound, in the basement membrane layers.
Thenumberofbasement membranebinding sitesis very
high
compared toIFN--y cellular receptor number(from 100to
10,000percell). This is well inexcessofthe numberof mole-cules neededto saturatecellularIFN--yreceptors(40). Further-more, theaffinity of IFN--y forbasement membrane is closeto
thereported
affinity
of this moleculefor its cellular receptor:10-9 to
10-"
M (40).Therefore,
we speculate that in vivo,basement membrane likely interacts with this
cytokine
and thatthiscytokinecould be inanactivestate.Inhibition experiments,
performed
toidentify
basement membrane components involved in IFN-ybinding,
showedTableII. DomainMappingofIFN--y
MAb Boundradioactivity
100 B27 92.1 B22 14.6
32 87.5
35 112.1
Radiolabeled IFN--y wasincubated overnightat4°Cwith heparan sulfate (I ,ug) dot-blottedontonitrocellulose,in thepresence ofthe
TableIII. Basement Membrane-bound IFN Antiviral Activity
Basement membrane-bound IFN Viable cells
ng %
0.379 64.0
0.193 57.7
0.055 38.8
0.016 13.0
0.004 11.3
0.001 5.7
0 5.1
Wish cells (5 *104/well)wereculturedovernightonmatrigel preincu-batedwithvarious quantitiesof
IFN-'y,
and thenchallenged for24 hwithvesicularstomatitis virus. Cell viability was determined using the3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay.Matrigel-boundIFN wasestimated in parallelexperimentswith radiolabeled IFN.
that heparan sulfatewas aneffective competitor.
Electropho-reticand dot-blotanalysisconfirmedthespecificity ofthe bind-ing.Heparansulfateproteoglycanisthebinding site for IFN-'y,
asin thecaseof basic FGF,
IL-3,
andGM-CSF(28,
39).
This isalso supportedby the fact thatheparinaseor
heparitinase
diges-tion of basementmembrane-glycosaminoglycan
significantly
reduced thebinding of
IFN-y.
The affinity of IFN-,y for heparan sulfate proteoglycan found here, 1.5 nM, ishigher thanthe
affinity
of basicFGF,610 nM(28), although basicFGF(pl=9.8) ismorebasicthan
IFN-y
(pl = 8.7). This suggests that thebinding ofIFN tobasement membraneisnotbasedon asimple electrostatic
in-teraction,
but may involveadefined region ofthemolecule. Moreover, heparansulfate
cannotsignificantly
fixribonucle-ase A (another basicprotein pl = 8.9) in our electrophoretic
system, showingthatthepresence ofa netpositive chargeon
the molecule isnot sufficient for
binding.
On the other side,IFN--y
doesnotbind
toothersulfated glycosaminoglycans suchasdermatan, keratan, orchondroitin sulfate, in spite of their negativecharge.Thus,this interaction appears to bespecific.
Thecarboxy-terminal partoftheIFN-ymolecule contains a clusterof nine amino acids:
Ala'24-Ser'32
(41) veryclosely relatedto a consensus sequence present in more than20knownheparin-binding proteins(42).Bindinginhibition experiments
have beendonewith fourMAbsdirectedagainst differentparts
of the IFN-y moleculetodetermine the
binding
siteonIFN-y. Onlyantibodies
directed against thecarboxy-terminal
partof the moleculepreventthebinding,
whereastheothers donot.Our data and thatofotherinvestigators (28, 43)suggest that extracellularmatrixcomponents couldprovidealocal
concen-tration ofagivencytokine, directthe rangeof action ofsuch soluble
factors,
andactas aphysiological
storagedepotaround cells. In the caseof basic FGF, itwassuggestedthat theregula-tionof capillarygrowth and the neovascular response may re-sult from displacement of growth factors from their storage siteswithinthebasement membrane and
surrounding
extracel-lularmatrix(28,44).Extracellularmatrix-heparan sulfatealsobindsIL-3 and GM-CSF. Oncebound, these cytokinescan be presented in an activeformtohemopoieticcells,thus provid-ingamechanistic explanation forthe observed absolute depen-dence of haemopoietic cells on intimatestromal cell contact (43).Interferon-'y hasanextremely broad range of biological
activities on a great number of different cells (1, 24). Once
boundtobasement membranes,thismolecule is still active.
Acknowledgments
Weare grateful to Dr. J. P. Escande and "SANOFI-Recherche" for the grantobtained and encouragement in this research. Dr. J. Banchereau (Laboratory for Immunological Research, Schering-Plough) is also
gratefullyacknowledged for his kindgiftof MAbs.
This work was supported by afellowship (H. Lortat-Jacob) from the Societe d'HepatologieExperimentale.
References
1. Balkwill, F. R., and F. Burke. 1989. Thecytokine network. Immunol. Today. 10:299-304.
2. Rosengurt, E., and M. Collins. 1983. Molecular aspects of growth factor action: receptors and intracellular signals. J. Pathol. 141:309-33 1.
3. Hay, E. D. 1981. Extracellular matrix. J. Cell Biol. 91:205s-223s.
4.Bissell,M. J., andM. H.Barcellos-Hoff. 1987. The influence of extracellu-larmatrix on gene expression: Is the structure the message? J. Cell Sci. Suppl. 8:327-343.
5. Ruoslahti, E., and M. D.Pierschbacher. 1987. New perspectives in cell adhesion: RGD and integrins. Science (Wash. DC). 238:491-497.
6.Juliano, R. L. 1987. Membrane receptors forextracellular matrix
macro-molecules:relationship to celladhesion and tumor metastasis. Biochim. Biophys. Acda.907:261-278.
7.Kleinman, H. K., R. J. Klebe, and G. R. Martin. 1981. Role of collagenous matrices in the adhesion and growth of cells. J. Cell Biol. 88:473-485.
8. Herbst, T. J., J. B. McCarthy, E.C. Tsilibary, and L. T. Furch. 1988. Differential effects of laminin, intact type IV collagen, and specific domains of type IVcollagen on endothelial cell adhesion and migration. J. Cell Biol.
106:1365-1373.
9. Duband, J. L., S.Dufour, and J. P. Thiery. 1988. Extracellular matrix-cy-toskeleton interactions in locomoting embryonic cells. Protoplasma.
145:112-119.
10.Kleinman, H. K., R. C. Ogle, F. B. Cannon, C. D. Little, T. M. Sweeney, and L.Luckenbill-Edds. 1988. Laminin receptorsfor neurite formation. Proc. Natl. Acad. Sci. USA. 85:1282-1286.
11.Watt, F. M. 1986. The extracellular matrix and cell shape. Trends Bio-chem. Sci. 11:482-485.
12.Liotta, L. A. 1986. Tumor invasion and metastases: role of the extracellu-lar matrix.Cancer Res. 46:1-7.
13.Liotta, L. A., C. N. Rao, and U. M. Wewer. 1986. Biochemical interac-tions of tumor cells with the basement membrane. Annu. Rev. Biochem. 55:1037-1057.
14.Pauli, B. U., and W. Knudson. 1988. Tumor invasion: a consequence of destructive andcompositional matrix alterations. Human Pathol. 19:628-639.
15.Morrison-Graham, K., and J. A. Weston. 1989. Mouse mutants provide newinsights into the role of extracellular matrix in cell migration and differentia-tion. Trends Genet. 5:116-121.
16.Lander, A. D., D. K.Fujii, D.Gospodarowicz, andL.F.Reichard. 1982. Characterization of a factor that promotes neuriteoutgrowth:evidencelinking
activity to heparan sulfate proteoglycan. J. Cell Biol. 94:574-585.
17.Kleinman, H. K.,F. B.Cannon, G.W.Laurie, J.R.Hassell,M.
Aumail-ley, V. P.Terranova, G. R.Martin, and M.Dubois-Dalcq. 1985.Biological activi-ties of laminin. J. Cell. Biochem. 27:317-325.
18.Chakravart, S., M. F. Tam, andA. E.Chung. 1990. The basement
mem-braneglycoproteinentactin promotes cell attachment and binds calciumions. J.
Biol. Chem. 265:10597-10603.
19.Timpl, R. 1989. Structure andbiologicalactivityof basement membrane proteins. Eur. J.Biochem. 180:487-502.
20. Kleinman, H.K., J.Graf,Y. Iwamoto, M.Sasaki,C. S.Schasteen,Y.
Yamada, G. R.Martin, and F.A.Robey. 1989. Identification ofasecondactive
site in lamininforpromotion of cell adhesion andmigrationand inhibition ofin
vivomelanoma lung colonization. Arch.Biochem.Biophys. 272:39-45. 21. Noonan, D.,E.Horigan, S.Ledbetter, G.Vogeli,M.Sasaki,Y.Yamada,
and J. R. Hassell. 1988.Identification ofcDNAclonesencodingdifferent
do-mains of the basement membrane heparan sulfate proteoglycan.J.Biol. Chem. 263:16379-16387.
22.Clement, B.,B.Segui-Real,J. R.Hassell, G.R.Martin,andY.Yamada. 1989. Identification of a cellsurface-binding protein for thecoreproteinofthe
basement membraneproteoglycan. J. Biol. Chem. 264:12467-12471. 23.Gallagher,J. T. 1989. The extendedfamilyofproteoglycans:social
resi-dentsof thepericellularspace. Curr.Opin. CellBiol. 6:1201-1208.
24.Trinchieri, G.,and B.Perussia. 1985. Immuneinterferon:apleiotropic
Iymphokinewith multipleeffects. Immunol. Today.6:131-136.
25. Clemens, M. J., and M. A.McNurlan. 1985. Regulation of cell
prolifera-tionanddifferentiationbyinterferons. Biochem. J. 226:345-360.
26.Shearer, M., and J. Taylor-Papadimitriou. 1987. Regulation ofcell growth byinterferons.Cancer Metastas. Rev. 6:199-221.
27. Romeo, G., G. Fiorucci, and G. B. Rossi. 1989. Interferons in cell growth anddevelopment. Trends Genet. 5:19-24.
28.Bashkin, P., S. Doctrow, M. Klagsbrun, C. M. Svahn, J. Folkman, and I. Vlodavsky. 1989. Basic fibroblast growth factor binds to subendothelial extracel-lular matrix and is released by heparitinase and heparin-like molecules. Biochem-istry.28:1737-1743.
29. Kleinman, H. K., M. L. McGarvey, J. R. Hassell, V. L. Star, F. B. Cannon, G. W.Laurie, and G. R. Martin. 1986. Basement membrane complexes with biological activity.Biochemistry.28:312-318.
30. Grant, D. S., H. K. Kleinman, C. P. Leblond, S. Inoue, A. E. Chung, and G.R.Martin. 1985. The basement membrane-like matrix of the mouse EHS tumor.II.Immunohistochemicalquantitation of six of its components. Am. J. Anat. 174:387-398.
31. Leblond, C. P., and S. Inoue. 1989. Structure, composition, and assembly of basement membrane. Am. J. Anat. 185:367-390.
32.Faltynek, C. R., G. L. Princler, and F. R. Ortaldo. 1986. Expression of IFN-alphaand IFN-gamma receptors on normal human small resting T lympho-cytes and large granularlymphocytes.J.Immunol. 136:4134-4139.
33.Chamness, G. C., and W. L. McGuire. 1975. Scatchard plot: common errors in correction and interpretation. Steroids. 26:538-542.
34.Stenberg, M., andH.Nygren. 1988. Kinetics of antigen-antibody
reac-tionsatsolid-liquid interfaces.J. Immunol. Methods. 113:3-15.
35.Cidadao,A.J. 1988.Interactions between fibronectin,
glycosaminogly-cans andnativecollagenfibrils:an EMstudy inartificialthree-dimensional extra-cellular matrices. Eur. J.CellBiol.48:303-312.
36. Favre, C., J.Wijdenes,H.Cabrillat,0.Djossou,J.Banchereau, and J. E. DeVries. 1989. Epitope mapping of recombinant human gamma interferon us-ing monoclonal antibodies. Mol. Immunol. 26:17-25.
37.Berg, K., M. B. Hansen, and S. E. Nielsen. 1990.A newsensitivebioassay forprecisequantification of interferonactivityasmeasured viathemitochondrial dehydrogenase function in cells(MTT-method).ActaPathol. Microbiol.
Im-munol.Scand.98:156-162.
38.Klotz, 1. M. 1982. Numbers of receptor site from Scatchard graphs: facts and fantasies. Science(Wash.DC). 217:1247-1249.
39. Burgess, A. W. 1988. Cell surface heparan sulfateimplicatedin
hae-mopoietic growth factor signalling.Immunol.Today.9:267-268.
40. Langer, J. A., and S. Pestka. 1988. Interferon receptors. Immunol.Today.
9:393-400.
41.Gray,P.W., D. W. Leung,D.Pennica,E.Yelverton, R.Najarian,C. C. Simonsen, R. Derynck,P.J.Sherwood,D. M.Wallace,S. L. Berger,etal.1982. Expression of human immune interferon cDNA inE. coli and monkey cells. Nature(Lond.).295:503-508.
42.Cardin, A. D., and H. J. R.Weintraub. 1989. Molecularmodelingof protein-glycosaminoglycan interactions.Arteriosclerosis.9:21-32.
43.Roberts, R. A., J. Gallagher,E.Spooncer,T.D.Allen,F.Bloomfield,and T. M.Dexter. 1988. Heparansulfateboundgrowth factors:amechanism for cell mediated haemopoiesis. Nature(Lond.). 332:376-378.
44.Baird, A., andN.Ling. 1987. FGFarepresent in the extracellular matrix produced by endothelial cells in vitro:implicationsforaroleofheparinase-like